diff --git a/server_stderr.txt b/server_stderr.txt
index 188148a..3582d34 100644
--- a/server_stderr.txt
+++ b/server_stderr.txt
@@ -58,3 +58,17 @@ Restored 1 datasets from disk
 [2026-07-24 12:30:18,840] INFO analysis.data_loader: [CLEAN] HEX columns added hex_pairs_count: ['0cph', '0crv', '0ver', '0rnt', '0rnd']
 [24/Jul/2026 12:30:37] "HEAD / HTTP/1.1" 200 0
 [24/Jul/2026 12:31:07] "HEAD / HTTP/1.1" 200 0
+[24/Jul/2026 12:31:37] "HEAD / HTTP/1.1" 200 0
+[24/Jul/2026 12:32:07] "HEAD / HTTP/1.1" 200 0
+[24/Jul/2026 12:32:37] "HEAD / HTTP/1.1" 200 0
+[24/Jul/2026 12:33:07] "HEAD / HTTP/1.1" 200 0
+[24/Jul/2026 12:33:37] "HEAD / HTTP/1.1" 200 0
+[24/Jul/2026 12:34:07] "HEAD / HTTP/1.1" 200 0
+[24/Jul/2026 12:34:37] "HEAD / HTTP/1.1" 200 0
+[24/Jul/2026 12:35:07] "HEAD / HTTP/1.1" 200 0
+[24/Jul/2026 12:35:37] "HEAD / HTTP/1.1" 200 0
+[24/Jul/2026 12:36:07] "HEAD / HTTP/1.1" 200 0
+[24/Jul/2026 12:36:37] "HEAD / HTTP/1.1" 200 0
+[24/Jul/2026 12:37:07] "HEAD / HTTP/1.1" 200 0
+[24/Jul/2026 12:37:37] "HEAD / HTTP/1.1" 200 0
+[24/Jul/2026 12:38:07] "HEAD / HTTP/1.1" 200 0
diff --git a/tianxuan/llm_orchestrator.py b/tianxuan/llm_orchestrator.py
index 09417c6..279ade7 100644
--- a/tianxuan/llm_orchestrator.py
+++ b/tianxuan/llm_orchestrator.py
@@ -314,8 +314,10 @@ def run_llm_pipeline(dataset_id: str, entity_col: str = "",
     # Track tool call history for smarter guidance
     called_tools: set[str] = set()
     consecutive_errors = 0
+    step = 0
+    max_steps = 999
 
-    for step in range(25):
+    while step < max_steps:
         resp = call_llm(msgs, config)
         if resp is None:
             if callback:
@@ -368,18 +370,9 @@ def run_llm_pipeline(dataset_id: str, entity_col: str = "",
             logger.info(f'[TOOL] step={step} name={name} args={json.dumps(args)[:200]}')
 
             try:
-                async def _call_with_timeout():
-                    return await asyncio.wait_for(
-                        handle_call(name, args),
-                        timeout=120,
-                    )
-                result = asyncio.run(_call_with_timeout())
-            except asyncio.TimeoutError:
-                result = {'error': f'Tool {name} timed out after 120s'}
-                logger.warning(f'[TOOL] {name} timed out')
+                result = asyncio.run(handle_call(name, args))
             except Exception as e:
                 result = {'error': f'MCP tool error: {e}'}
-                logger.error(f'[TOOL] {name} error: {e}')
 
             if callback:
                 callback(step, name, result, meta={"args": args})
@@ -417,6 +410,8 @@ def run_llm_pipeline(dataset_id: str, entity_col: str = "",
                     "content": f"Tool '{name}' failed: {result['error'][:200]}. Call '{suggested}' to diagnose, then retry with corrected parameters.",
                 })
 
+        step += 1
+
     if callback:
-        callback(25, "max_steps", None)
-    return {"error": "Reached max 25 tool calls", "steps": 25}
+        callback(max_steps, "max_steps", None)
+    return {"error": f"Reached max {max_steps} tool calls", "steps": max_steps}